From dbe25904b91b26cb80e2d8383c127406bdec5b57 Mon Sep 17 00:00:00 2001
From: Kris <2893855659@qq.com>
Date: Wed, 8 Jul 2026 22:58:06 +0800
Subject: [PATCH] feat(wechat_woc): add delivered parts skip logic for
 multi-part messages
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

新增已发送分片跳过逻辑，支持重复发送时只发送未成功投递的消息分片，避免重复推送已送达内容
---
 .../wechat_woc/adapters/outbound_adapter.py   | 30 +++++++++++--------
 1 file changed, 18 insertions(+), 12 deletions(-)

diff --git a/backend/package/yuxi/channels/plugins/wechat_woc/adapters/outbound_adapter.py b/backend/package/yuxi/channels/plugins/wechat_woc/adapters/outbound_adapter.py
index ad08def3..a2299d2c 100644
--- a/backend/package/yuxi/channels/plugins/wechat_woc/adapters/outbound_adapter.py
+++ b/backend/package/yuxi/channels/plugins/wechat_woc/adapters/outbound_adapter.py
@@ -252,6 +252,7 @@ class WeChatWocOutboundAdapter:
         *,
         idempotency_key: str | None = None,
         channel_request_id: str | None = None,
+        delivered_parts: list[int] | None = None,
     ) -> list[MultiPartReceipt]:
         """发送多部分消息（文本在前，媒体在后）。
 
@@ -306,6 +307,7 @@ class WeChatWocOutboundAdapter:
         """
         receipts: list[MultiPartReceipt] = []
         seq = 0
+        skip_seqs = set(delivered_parts) if delivered_parts else set()
 
         # 1. 发送前先校验所有媒体附件的出站能力，避免文本已发后才发现媒体
         #    不支持导致文本无法回滚（M-3 修复）。任一媒体不支持即整体失败，
@@ -342,23 +344,27 @@ class WeChatWocOutboundAdapter:
 
         # 2. 文本在前
         if payload.content:
-            resp = await self._client.send_text(
-                account_id,
-                peer_id,
-                payload.content,
-            )
-            receipts.append(
-                MultiPartReceipt(
-                    platform_msg_id=_extract_channel_msg_id(resp),
-                    part_type=_PART_TYPE_TEXT,
-                    sequence=seq,
-                    success=True,
+            if seq not in skip_seqs:
+                resp = await self._client.send_text(
+                    account_id,
+                    peer_id,
+                    payload.content,
+                )
+                receipts.append(
+                    MultiPartReceipt(
+                        platform_msg_id=_extract_channel_msg_id(resp),
+                        part_type=_PART_TYPE_TEXT,
+                        sequence=seq,
+                        success=True,
+                    )
                 )
-            )
             seq += 1
 
         # 3. 媒体在后（图片 / 文件）
         for attachment in payload.attachments:
+            if seq in skip_seqs:
+                seq += 1
+                continue
             # attachment.url 必须为 bridge 容器内可访问的绝对路径（见方法
             # docstring 中 ``attachment.url`` 语义重载说明）
             file_path = attachment.url